T-Cell Infiltration and Adaptive Treg Resistance in Response to Androgen Deprivation With or Without Vaccination in Localized Prostate Cancer

Abstract

Purpose: Previous studies suggest that androgen deprivation therapy (ADT) promotes antitumor immunity in prostate cancer. Whether a vaccine-based approach can augment this effect remains unknown.

Patients and methods: We conducted a neoadjuvant, randomized study to quantify the immunologic effects of a GM-CSF-secreting allogeneic cellular vaccine in combination with low-dose cyclophosphamide (Cy/GVAX) followed by degarelix versus degarelix alone in patients with high-risk localized prostate adenocarcinoma who were planned for radical prostatectomy.

Results: Both Cy/GVAX plus degarelix and degarelix alone led to significant increases in intratumoral CD8⁺ T-cell infiltration and PD-L1 expression as compared with a cohort of untreated, matched controls. However, the CD8⁺ T-cell infiltrate was accompanied by a proportional increase in regulatory T cells (Treg), suggesting that adaptive Treg resistance may dampen the immunogenicity of ADT. Although Cy/GVAX followed by degarelix was associated with a modest improvement in time-to-PSA progression and time-to-next treatment, as well as an increase in PD-L1, there was no difference in the CD8⁺ T-cell infiltrate as compared with degarelix alone. Gene expression profiling demonstrated that CHIT1, a macrophage marker, was differentially upregulated with Cy/GVAX plus degarelix compared with degarelix alone.

Conclusions: Our results highlight that ADT with or without Cy/GVAX induces a complex immune response within the prostate tumor microenvironment. These data have important implications for combining ADT with immunotherapy. In particular, our finding that ADT increases both CD8⁺ T cells and Tregs supports the development of regimens combining ADT with Treg-depleting agents in the treatment of prostate cancer.

Figure 1.. Clinical trial design and patient disposition diagram.

Patients with high-risk localized prostate cancer (T1c–3b N0 M0, Gleason 7–10) were randomized 1:1 to degarelix (240 mg SQ) vs. Cyclophosphamide (200 mg/m2 IV) / GVAX (2.5×108 PC3 cells, 1.6×108 LNCaP cells) given 2 weeks before degarelix. All patients then underwent radical prostatectomy 2 weeks after degarelix. Abbreviations: Eastern Cooperative Oncology Group (ECOG); subcutaneously (SQ); intravenously (IV).

Figure 2.. Degarelix and degarelix + GVAX increase CD8+ and FOXP3+ T cell infitration in prostate tumors.

A) Representatitive H&E and immunohistochemistry (IHC) for CD8+ T cells, visualized at 4x and 20x magnification B) Representative H&E and IHC for FOXP3+ T cells, visualized at 4x and 20x magnification C) Boxplots of Log2(CD8+ T cell density), quantified from IHC as represented in Figure 2A. D) Boxplots of Log2(FOXP3+ T cell density), quantified from IHC as represented in Figure 2B. E) Boxplots of the CD8+/FOXP3+ T cell ratio, quantified from IHC as represented in Figures 2A and 2B. F) Table of mean CD8+ T cell density (cells/mm²), mean Treg density (cells/mm²), and CD8/Treg ratio for each treatment group and untreated controls, with 95% confidence intervals and p-values by Gleason-stratification-adjusted ANOVA reported for each comparison of groups; * = p<0.05, ** = p<0.01, and *** = p<0.005

Figure 3.. Degarelix and degarelix + Cy/GVAX increase PD-L1 expression in prostate tumors.

A) Representative IHC for PD-L1, visualized at 4x and 20x magnification B) Stacked barplot of %PD-L1 positive cells, showing relative proportion of samples with 0% PD-L1 staining, <1% PD-L1 staining, <5% PD-L1 staining, and >5% PD-L1 staining in tumor cells in each treatment group and a cohort of untreated matched controls. Distributions of %PD-L1 categories may be visually compared between groups, such that the degarelix + Cy/GVAX group has the highest proportion of samples with PD-L1 > 5%. Proportions of samples with %PD-L1 > 0 were also compared between groups by Fisher’s exact test, with p-values shown above the plot for each comparison, where * = p<0.05, ** = p<0.01, and *** = p<0.005.

Figure 4.. Degarelix and degarelix + Cy/GVAX induce complex changes in immune gene signatures in primary prostate tumors.

A) Differential expression of immune related genes by Nanostring Immune Profiling Panel in primary prostate tumors after degarelix, degarelix + Cy/GVAX, and untreated matched controls. Euler plots showing number of genes with Benjamini-Hochberg corrected t-test p-value < 0.01 for each pairwise comparison of groups, such that “up-regulated genes” refers to genes that have higher mean frequency in the degarelix + Cy/GVAX group than in the degarelix group (cyan), higher mean frequency in the degarelix group than the untreated control group (purple), and higher mean frequency in the degarelix + Cy/GVAX group than the untreated control group (orange), and “down-regulated genes” refers to genes that have lower mean frequency in degarelix + Cy/GVAX vs degarelix (cyan), degarelix vs controls (purple), and degarelix + Cy/GVAX vs controls (orange), respectively. B) Violin-plot of log-scaled post-normalization Nanostring gene counts for CHIT1 in each treatment group and untreated controls. In Figure 4A, CHIT1 is the sole gene significantly up-regulated in each comparison. C) Boxplot of immune cell type absolute abundances as inferred by CIBERSORT, colored by treatment group and reported for all samples with CIBERSORT p-value<0.05. D) Violin-plot of total immune cell infiltrate for each sample by treatment group, such that total immune cell infiltrate represents the sum of CIBERSORT immune cell abundances as shown in Figure 4C E) Boxplot of immune cell populations for which t-test comparing abundance between groups showed an uncorrected p-value<0.05. P-values were obtained by unpaired t-test with Benjamini-Hochberg multiple-testing correction, and shown on Figures 4B and 4D with * = p<0.05, ** = p<0.01, and *** = p<0.005.

Figure 5.. Combination of Cy/GVAX with degarelix improves time-to-PSA recurrence and increases time-to-next treatment.

A) Kaplan-Meier curves comparing time-to-PSA recurrence of patients treated with degarelix + Cy/GVAX vs degarelix alone. Informative clinical variables for multivariate analysis were selected by backward feature selection using the Akaike Information Criterion. B) Forest plot showing time-to-PSA recurrence hazard ratios with 95% confidence interval for multiple cox regression of progression-free-survival against Cy/GVAX status, patient age, tumor stage, and Gleason score. P-values for each variable are reported, as is the overall log-rank p-value, Akaike Information Criterion value, and concordance index for the regression C) Kaplan-Meier curves comparing time to next treatment for patients treated with degarelix + Cy/GVAX vs degarelix alone, with log-rank p-value reported from multiple cox regression of time-to-next-treatment against Cy/GVAX status, patient age, tumor stage, and Gleason score. Informative clinical variables were selected as in 5A D) Forest plot showing time-to-next-treatment hazard ratios with 95% confidence interval for multiple cox regression of time-to-next-treatment against Cy/GVAX status, patient age, tumor stage, and gleason score. P-values for each variable are reported, as is the overall log-rank p-value, Akaike Information Criterion value, and concordance index for the regression.